And the ex-statistician jumps in again to note that, in the 51-patient severe/critical trial "sneak peak," if (a) at least of 9 of 17 initial placebo patients die and (b) the death rate in the 34 person leronlimab group is half or lower than the placebo group, the significance level (p-value) will be strong enough to argue for early termination.
IMO, assuming the usual formula for difference in binomials. Actual randomization may not be 34/17.